💻 computer science

Non-classical Topological Evidence Logic

This paper demonstrates that Topological Evidence Logic (TEL) is robust under modifications to its propositional base by extending the framework to intuitionistic and relevant logics, ultimately establishing a sound and complete system for relevant TEL based on the weak relevant modal logic BS4.

Original authors: Igor Sedlár (Institute of Computer Science, Czech Academy of Sciences)

Published 2026-07-01

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Igor Sedlár (Institute of Computer Science, Czech Academy of Sciences)

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: How We Know What We Know

Imagine you are trying to figure out if a hypothesis is true. In the world of logic, we usually have a "toolbox" of rules to help us decide. For a long time, logicians used a very strict, black-and-white toolbox called Classical Logic. It assumes that every statement is either 100% true or 100% false, and that if you know one thing, you automatically know everything that follows from it.

However, the author, Igor Sedlár, argues that real people (and even some computer systems) don't reason like that. We often deal with incomplete information, contradictions, or shades of gray. This paper tries to build a new kind of "evidence logic" that works better for these messy, real-world scenarios.

The Original Idea: The "Dense" Map

The paper starts with a concept called Topological Evidence Logic (TEL). To understand this, imagine you are a cartographer trying to map a territory.

The Map (Topology): Instead of drawing every single tree, you draw "open areas" where you have confirmed evidence.
Coherent Justification: In the original TEL, a hypothesis is considered "coherently justified" (or truly known) if it is supported by a dense open set.
- The Analogy: Imagine you are looking for a specific type of flower in a meadow. You don't need to see the flower in every square inch of the meadow. You just need to find a patch of evidence (an "open set") that is so widespread ("dense") that no matter where you look in the meadow, you are guaranteed to be close to that patch. If your evidence covers the meadow so thoroughly that you can't avoid it, then your hypothesis is justified.

The original version of this logic worked great, but it relied on the strict "black-and-white" Classical Logic toolbox. The paper asks: What happens if we change the toolbox to one that handles gray areas or contradictions?

Part 1: The Intuitionistic Version (The "Maybe" Logic)

First, the author tries Intuitionistic Logic. Think of this as a logic of "building knowledge." In this world, you can't just say "It's not true" unless you have a proof that it's impossible. It's like a construction site: you can't say a wall is "finished" until you've actually built it.

The Challenge: The original TEL needed a specific tool (Boolean negation) to define "density." In the "building" logic, that specific tool doesn't exist in the same way.
The Solution: The author shows that if you add a special "Global Modality" (a tool that lets you look at the entire map at once, not just your current spot), you can still define "density."
The Result: You can successfully build a version of TEL that works with "building" logic. The logic remains robust; it just needs a slightly different set of instructions to handle the "maybe" nature of the evidence.

Part 2: The Relevant Version (The "Connected" Logic)

Next, the author tries Relevant Logic. This is the most interesting part. In classical logic, if you believe "The moon is made of cheese," you might accidentally be forced to believe "The moon is made of cheese, therefore I am the King of France" (because in strict logic, a false premise can prove anything).

Relevant logic says: No! Your conclusion must actually be connected to your premise. If the premise has nothing to do with the conclusion, the argument is invalid. It's like a conversation where you can't suddenly jump to a completely unrelated topic without a bridge.

The Failure: The author first tries to use the standard Relevant Logic toolbox with the original TEL rules. It fails.
- Why? In this "connected" world, the standard tools can't express the idea of "interior of a complement" (which is needed to define density). It's like trying to measure the empty space inside a box using only a ruler that measures solid objects. The math breaks down; you can't prove that your evidence is "dense" enough.
The Fix: The author invents a new tool for the toolbox: an "Interior-of-Complement" operator.
- The Analogy: Imagine you have a flashlight (the standard tool) that shows you what is inside a room. The new tool is a "Shadow-Flashlight" that shows you the shape of the empty space outside the room. By adding this new tool, the logic can finally "see" the empty spaces and calculate density correctly.
The Result: With this new tool added, the author successfully creates a Relevant Topological Evidence Logic. They prove that this new system is sound (it doesn't produce nonsense) and complete (it can prove everything it's supposed to).

The Main Takeaway

The paper is a technical demonstration of robustness.

Think of the original Topological Evidence Logic as a house built on a specific type of concrete (Classical Logic). The author asks: "If we change the concrete to something softer (Intuitionistic) or something that requires different structural beams (Relevant), does the house fall down?"

Answer: No, the house stands firm.
How?
- For the "soft" concrete, we just needed to add a global view (the Global Modality).
- For the "structural" concrete, we had to invent a new tool (the Interior-of-Complement operator) to make the math work.

The paper concludes that the idea of "coherent justification" (knowing something because your evidence is everywhere) is a powerful concept that can survive even when we change the fundamental rules of how we reason. It doesn't break; it just needs to be adapted.

Technical Summary: Non-classical Topological Evidence Logic

Problem Statement
Topological Evidence Logic (TEL) is an epistemic framework where a hypothesis is considered "coherently justified" if and only if it is entailed by a dense open set within a topological space. In classical TEL, this is typically formulated as an extension of the modal logic S4 with a global modality ( $A$ ), where the interior operator ( $\square$ ) represents evidence. However, all existing formulations of TEL rely on classical propositional logic. The author notes that classical logic has been criticized for misrepresenting ordinary agent reasoning, particularly regarding closure principles (e.g., believing the consequent of an implication implies believing the implication). While non-classical epistemic logics (intuitionistic and relevant) have been developed to address these issues, they currently lack the topological machinery to express "coherent justification" via dense open sets. The central problem addressed is whether the TEL approach is robust under modifications to the propositional base, specifically within intuitionistic and relevant logic frameworks.

Methodology
The paper employs topological semantics over relational structures to extend TEL into non-classical settings.

Intuitionistic TEL: The author extends the work of de Groot and Shillito on the intuitionistic modal logic iS4. The methodology involves defining "up-spaces" (pre-ordered sets equipped with an up-set topology) and introducing a global modality ( $A$ ). The satisfaction relation is defined such that $\square\phi$ corresponds to the interior of the set of worlds satisfying $\phi$ , and $A\phi$ holds if $\phi$ is true in all worlds. The author demonstrates that despite the non-classical nature of negation, the combination of the interior operator and intuitionistic negation allows for the expression of the interior of a complement, which is necessary to define density.
Relevant TEL: The author investigates the application of TEL to relevant logic, specifically using Routley–Meyer frames extended with an up-set topology (RM up-spaces). A naive extension of the language (adding $\square$ and $A$ to the basic relevant logic $B$ ) is shown to be insufficient. Through counter-models, the paper demonstrates that in this restricted language, one cannot express the interior of a complement or the existence of a dense open set supporting a proposition. To resolve this, the language is extended with a new operator, $\square_c$ , interpreted as the "interior of the complement" ( $Int(P^c)$ ). This extension allows for the definition of density and coherent justification within the relevant setting.
Axiomatization: For both systems, the paper constructs Hilbert-style axiomatizations. The completeness proofs utilize canonical model constructions. A specific challenge in modal relevant logic is the interaction between the global modality and the canonical model; the author employs an "anchoring" technique, restricting the canonical model to theories that agree with a fixed "anchor" theory on all global formulas ( $A\phi$ ).

Key Contributions and Results

Intuitionistic TEL: The paper proves that coherent justification can be expressed in an extension of de Groot and Shillito's iS4 with a global modality $A$ . It provides a sound and complete axiomatization (system iS4A) for this logic. The key technical insight is that intuitionistic negation combined with the interior operator suffices to define the interior of a complement, thereby enabling the expression of density ( $A\diamond\square\phi$ ).
Relevant TEL: The paper establishes that a standard relevant modal logic with a global modality is insufficient for TEL. It introduces BS4A,c, a logic extending the weak relevant modal logic BS4 with a global modality $A$ and the interior-of-complement operator $\square_c$ .
- It proves that in BS4A,c, coherent justification is expressible as $\square\phi \land A\diamond_c\square\phi$ (where $\diamond_c$ is the dual of $\square_c$ ).
- It provides a sound and complete axiomatization for BS4A,c.
- As a byproduct, this result solves an open problem regarding the axiomatization of relevant logic with a global modality ( $A$ ) left open in previous work (Standefer et al.), as the axiomatization of BS4A,c (minus the modal axioms for $\square$ and $\square_c$ ) yields a complete axiomatization for the basic relevant logic with $A$ .
Expressivity Limits: The paper rigorously demonstrates via counter-models (Figures 2, 3, and 4) that without the $\square_c$ operator, relevant logic cannot express the interior of a complement or the property of a set being supported by a dense open set. Furthermore, it shows that even with $\square_c$ , the language cannot express intuitionistic implication.

Significance and Claims
The paper claims to bridge the gap between topological evidence-based epistemic logic and non-classical propositional logics. Its primary significance lies in demonstrating the robustness of the TEL approach: the concept of coherent justification, defined topologically, can be successfully transferred to intuitionistic and relevant settings, provided the language is appropriately extended.

The author modestly characterizes the work as "the first steps" in this direction. The paper does not claim to resolve all issues in non-classical epistemic logic but rather establishes the foundational semantic and syntactic frameworks for intuitionistic and relevant TEL. The main technical achievement is the soundness and completeness theorem for BS4A,c, which validates the use of topological tools in weak relevant modal logics. The paper concludes by identifying future work, including extending decidability results to the new systems, investigating stronger relevant logics (like R or E), and exploring the relationship between topological spaces and quantales in the context of RM frames.

The Big Picture: How We Know What We Know

The Original Idea: The "Dense" Map

Part 1: The Intuitionistic Version (The "Maybe" Logic)

Part 2: The Relevant Version (The "Connected" Logic)

The Main Takeaway

More like this